Objectives
Lecture 6 · p.2- What is hypothesis?
- What is testing a hypothesis?
- Why do you need statistical tests to test hypothesis?
- Which statistical tests do I use for testing hypothesis?
Slide errata
3L6 p.23 and Reality Check 2 (p.66) · Tension with standard theory
The slides say 'Alpha error decreases as the sample size increases' and mark L6 RC2 ('You can decrease alpha error by increasing the sample size') True. In standard theory alpha is fixed by the chosen threshold (e.g. 5%) whatever n is; a larger n reduces beta.
what the exam expects:True: alpha error can be decreased by increasing the sample size (the module's answer).
How the app teaches it: Teach the module's answer for exam purposes, and flag that in standard theory α is fixed by the threshold and n reduces β.
Alpha error · Beta error and power
L6 p.15 · Slide error
The alternate to null hypothesis 2 ('the difference is less than 5%') is given as 'is not 5%'. The logical opposite of '< 5%' is '≥ 5%'.
what the exam expects:H₀: the difference is less than the MCID; H₁: the difference is at least the MCID.
How the app teaches it: Use '≥ 5%' (at least 5%). This matches EQUAL-STATS in the PBL Session 2 answer.
Test, null and alternate hypotheses
Mock Q24 · Unconfirmed answer key
There is no official mock answer key, and no slide states the non-inferiority null explicitly (L9 p.31 gives only the margin and one-sided alpha). D ('increase in pain score … more than 2 cm') is the most likely answer, but the stem says '2 cm (or more)', so a strict reading could favour E.
what the exam expects:D (most likely; unconfirmed).
How the app teaches it: Show D as the answer with a 'check with your tutor' flag until the course team confirms it.
One- vs two-sided; equivalence